Fix WINDOWS32 bug #11155 with patch from Alessandro Vesely.

This commit is contained in:
Paul Smith 2004-11-28 16:58:51 +00:00
parent e8e4c19873
commit d27cac1598
2 changed files with 13 additions and 4 deletions

View file

@ -1,3 +1,9 @@
2004-11-28 Paul D. Smith <psmith@gnu.org>
* main.c (find_and_set_default_shell) [WINDOWS32]: check for
equality of "cmd"/"cmd.exe", not inequality. Fixes bug #11155.
Patch by Alessandro Vesely.
2004-11-12 Paul D. Smith <psmith@gnu.org>
* job.c (child_execute_job) [VMS]: Don't treat "#" as a comment on

11
main.c
View file

@ -718,11 +718,11 @@ find_and_set_default_shell (char *token)
if (((tokend == search_token
|| (tokend > search_token
&& (tokend[-1] == '/' || tokend[-1] == '\\')))
&& strcmpi (tokend, "cmd"))
&& !strcmpi (tokend, "cmd"))
|| ((tokend - 4 == search_token
|| (tokend - 4 > search_token
&& (tokend[-5] == '/' || tokend[-5] == '\\')))
&& strcmpi (tokend - 4, "cmd.exe"))) {
&& !strcmpi (tokend - 4, "cmd.exe"))) {
batch_mode_shell = 1;
unixy_shell = 0;
sh_found = 0;
@ -1071,8 +1071,11 @@ main (int argc, char **argv, char **envp)
We used to rely on target_environment's v_default code to do this.
But that does not work for the case where an environment variable
is redefined in a makefile with `override'; it should then still
be exported, because it was originally in the environment. */
->export = v_export;
be exported, because it was originally in the environment.
Another wrinkle is that POSIX says the value of SHELL set in the
makefile should not change the value of SHELL given to
subprocesses, which seems silly to me but... */
->export = strncmp(envp[i], "SHELL=", 6) ? v_noexport : v_export;
}
#ifdef WINDOWS32
/*